面向植被覆盖的土地利用类型分类方法
Land Use Type Classification Method for Vegetation Coverage

随着高光谱遥感图像和神经网络的应用，基于高光谱遥感图像的植被覆盖类型分类进一步的发展。然而由于高光谱遥感图像具有更加丰富的光谱信息，传统分类方法不能很好地同时提取光谱特征和空谱特征。为了解决上述问题，本文以U-Net模型为框架，融合了3D-CNN和多尺度特征提取模块，提出了一种基于三维卷积和多尺度特征融合的神经网络模型，使得网络在进行特征提取和分类的过程中，更好的融合光谱信息、空间信息、全局信息和细节信息，最终使模型的分类准确度得以提升。
With the application of hyperspectral remote sensing images and neural networks, the classification of vegetation cover types based on hyperspectral remote sensing images has been further developed. However, since hyperspectral remote sensing images have more abundant spectral information, traditional classification methods cannot extract spectral features and spatial-spectral features at the same time. In order to solve the above problems, this paper takes the U-Net model as the framework, integrates the 3D-CNN and multi-scale feature extraction module, and proposes a neural network model based on three-dimensional convolution and multi-scale feature fusion. In the process of feature extraction and classification, the network better integrates spectral information, spatial information, global information and detail information, and finally improves the classification accuracy of the model.